Conventionally, in digital still cameras, video cameras, and so forth, in order to prevent image degradation due to cancellation errors in digital signal processing, the grayscale depth of the image signal in the input and processing systems (around 10 to 12 bits) is set to be large compared to the grayscale depth of the final output signal (normally 8 bits). In this case, it is necessary to perform grayscale conversion so as to match the grayscale depth of the output system. Generally, grayscale conversion is performed with static grayscale characteristics for a reference image scene. However, when generating an image signal having grayscale characteristics that are subjectively preferable like the human visual function, one approach that is known to be suitable is to perform different grayscale conversion in each region in the image (hereinafter called space-variant grayscale conversion).
For example, the Publication of Japanese Patent No. 2951909 discloses an example in which the image signal is divided into a plurality of regions on the basis of the luminance distribution of the input image and optimum grayscale conversion is performed for each region to match the photographic subject conditions. In addition, the Publication of Japanese Patent No. 3465226 discloses an example in which the image signal is divided into regions on the basis of texture information, and suitable grayscale conversion is performed in each region.
Patent Document 1: Publication of Japanese Patent No. 2951909.
Patent Document 2: Publication of Japanese Patent No. 3465226